Machine learning developed LKB1-AMPK signaling related signature for prognosis and drug sensitivity in hepatocellular carcinoma.

Hepatocellular carcinoma (HCC) is one of the most common tumors worldwide, posing a significant threat to the life and health of people globally. LKB1-AMPK signaling pathway plays a significant role in the regulation of cellular metabolism, proliferation and survival in cancer. To construct a LKB1-AMPK signaling-related gene signature (LRS), an ensemble of ten machine learning algorithms was applied across four datasets. Several indicators were employed to assess the effectiveness of LRS in forecasting immunological responses. Additionally, in vitro studies were conducted to investigate the biological roles of LKB1 in HCC. The optimal LRS developed using the Lasso algorithm served as a significant risk factor for HCC patients. HCC patients with a high LRS score exhibited poorer prognoses, with 1-, 3-, and 5-year ROC AUC values of 0.863, 0.826, and 0.831, respectively. Conversely, a low LRS score was associated with higher levels of CD8(+) T cells, NK cells, macrophages M1, cytolytic activity, T cell co-stimulation and ESTIMATE scores. Additionally, HCC cases with lower LRS score showed elevated PD1&CTLA4 immunophenoscores, and TMB scores, while exhibiting reduced TIDE and tumor escape scores. The IC50 values for several chemotherapy and targeted therapy were found to be lower in HCC cases with higher LRS score. Furthermore, gene set enrichment analysis revealed that pathways related to angiogenesis and NOTCH signaling were more active in the high LRS score group. Over-expression of LKB1 led to decreased proliferation, migration, and invasion in HCC cells by regulating AMPK and PD-L1 expression. Our investigation developed a novel LRS for HCC, serving as an indicator for predicting clinical outcome and immunotherapy response.